EmergingConversation, prompt and context design

Prompt Engineer

Also called Prompt Designer

A Prompt Engineer writes and tests the instructions and examples that make a model do one task well. They turn vague prompting into versioned, measured work that holds up when models and inputs change.

What they do

They take a task, such as classifying support requests or drafting structured summaries, and write instructions and examples for a model to follow. They build a small set of test cases, including awkward ones, and keep some back as held-out data. They run variations, read the failures closely and sort them into causes. Then they change the prompt, record the version and rerun the tests. They also check that a prompt still works when the model or the input style changes. Much of the week is reading outputs and writing down what went wrong.

What a strong hire delivers

In the first months, a strong hire gives the team a versioned prompt library for its most important tasks, each with a held-out evaluation set. Prompt changes stop being guesswork because every change is tested and logged. The team gets a clear view of where models fail on its tasks. Prompts survive a model switch with known, measured effects. Engineers and product staff can reuse the library without asking the author.

How you can tell they are good

Ask them to walk through a prompt they improved. A good candidate shows the failure cases first, explains how they grouped them, and shows what changed and what the held-out results said. They can say what did not work. Ask how they would know a gain is real. Be wary of anyone who talks only about clever phrasing, or who judges prompts by a handful of good-looking examples.

Why it is hard to fill

The title is inconsistent. The work often sits inside conversation design, content design, linguistics or product roles, so CVs rarely say Prompt Engineer. Many people who use the title do informal tinkering without evaluation discipline. The scarce person combines language skill with experimental method and enough technical comfort to work with engineers. Hiring on title alone misses most of them, and hiring on enthusiasm alone finds the wrong ones.

Where to find them

Look at conversation designers, content designers, linguists, localisation specialists and application designers who have moved into model-facing work. Many are in product or design teams at companies building assistants, support automation or search features. Others sit in applied research or engineering teams, where prompting is part of a wider job. Ask about the work they do rather than the title they hold.

Indicative pay

Not enough published evidence yet to give a fair range for this role, which is common for work this new. We benchmark it for you as part of a search.

Interview questions for a Prompt Engineer

  1. Describe a prompt you improved. What was failing, and how did you find out?

    A strong answer starts with observed failures, shows how they were grouped and tested, and names the measured result. Worry if the story is about intuition, wording tricks or a few hand-picked examples.

  2. How do you build a held-out evaluation set for a new task with no existing data?

    Look for coverage of typical and awkward inputs, clear pass criteria, and keeping some cases unseen while tuning. Worry if they tune and test on the same examples or ignore how inputs vary.

4 more questions, with what a strong answer covers, are in the Role Kit.

Job description you can post

We are hiring a Prompt Engineer to make our model-powered features reliable on the tasks that matter to our customers. You will write and test instructions and examples, measure the results and leave behind work the whole team can reuse.

What you will do

  • Write and refine task-specific instructions and examples for model-powered features
  • Build and maintain evaluation sets, including held-out cases
  • Analyse failures and group them by cause
  • Keep a versioned prompt library with clear change records
  • Check prompts still hold up when models or inputs change
  • Work with engineers and designers to turn findings into product changes

What you will bring

  • Examples of prompts you have improved, with evidence of measured gains
  • Experience designing evaluations and reading model outputs critically
  • Strong written English and an eye for ambiguity
  • Background in conversation design, content design, linguistics or application design, or equivalent work
  • Comfort working with engineers and simple scripts or test tooling

Nice to have

  • Experience with multilingual or localised content
  • Experience with service design or support workflows
  • Familiarity with agent frameworks

Free to use and adapt. If it saves you time, a link back to this page is appreciated.

Skills that matter

  • Service and conversation designAdvancedStructure dialogue, escalation and service experiences
  • Prompt and context designAdvancedStructure instructions, examples and context for a task
  • Model evaluationWorkingMeasure capability and failure modes credibly
  • Linguistics and localisationWorkingAdapt language behaviour across cultural and linguistic contexts

Other roles in conversation, prompt and context design

Hiring nowConversation DesignerA Conversation Designer shapes how an AI service talks, asks, hands over and recovers. They turn model behaviour into dialogue people can finish…EmergingContext EngineerA Context Engineer decides what information, memory and tools reach the model at each step. That choice shapes answer quality, cost and privacy, so…Hiring nowAI Content DesignerAn AI Content Designer writes the words inside an AI product: prompts users see, explanations, warnings and error messages. Good content helps…

Related roles in other families

Hiring nowNatural Language Processing ScientistLanguage and multilingual intelligenceAn NLP Scientist builds and tests methods for language understanding and generation. They make sure a company's language models work across…Hiring nowMultilingual Model ResearcherLanguage and multilingual intelligenceA Multilingual Model Researcher improves how well a model transfers and behaves across languages. It matters because a product that works in one…Hiring nowLanguage Localisation EvaluatorLanguage and multilingual intelligenceA Language Localisation Evaluator judges whether model output means the right thing, sounds right and is appropriate in a specific locale. It…EmergingLow-Resource Language Data ResearcherLanguage and multilingual intelligenceA Low-Resource Language Data Researcher builds ethically sourced datasets and evaluations for languages that models serve poorly. It matters because…

How we wrote this

Role definitions are our own editorial work, checked against occupational classifications, standards, research and employer pages. Titles vary by company, and this is a description of the work, not a job advert.

← Conversation, prompt and context designAll new job families

Start a search

Hiring a Prompt Engineer?

Tell us the work. We find the people already doing it, whatever their job title says, and you meet them already interviewed for your role.